In order to obtain the spectral information of objects and improve the retrieval of quantitative parameters from remotely sensing data accurately on land or over water bodies, atmospheric correction is a vital step, certainly, it is also a prerequisite to hyperspectral imagery data analysis approaches. On the base of previous studies, the atmospheric correction algorithms were divided to two categories: image-based empirical and model-based correction methods. The Quick Atmospheric Correction (QUAC) and Dark Object Subtraction (DOS) methods belong to the empirical or semiempirical methods, however, the Fast Line-of-sight Atmospheric Analysis of Spectral Hypercube (FLAASH) method was developed from the radiative transfer model. In this paper, we initially evaluated the performance from Hyperspectral Imager for the Coastal Ocean (HICO) of 16 Nov 2013 using QUAC, DOS, and MODTRAN integrated in FLAASH, and compared the results of these correction methods with in situ data. The results indicate that the method of FLAASH model performs much better than DOS and QUAC in atmospheric correction for HICO hyperspectral imagery, although the DOS and QUAC method is conducted more easily and do not require inputs of complex parameters.